MACROECONOMIC DATA AND FORECASTS

Tara M. SinclairAssociate Professor of Economics, The George Washington University and Chief Economist, IndeedEmail: tsinc@gwu.eduWeb: http://home.gwu.edu/~tsinc/Twitter: @TaraSinc

Weidenbaum Center Media RetreatWianno Club in Cape CodJune 28, 2016

“An economist is an expert who will know tomorrow why the things he

predicted yesterday didn't happen today.”- Laurence J. Peter

The most-watched government data are old news…

and then they’re revised.

U.S. Real GDP Growth Estimates1st Quarter 2015

AdvanceEstimate4/29/15

0.2

-0.2

-0.7

0.6

SecondEstimate5/29/15

ThirdEstimate6/24/15

AnnualRevision7/30/15

~ 40% of advance numbers are from “trend” models!

Revisions Between Quarterly Annualized Percent Changes of Real GDP

Revision Average Absolute Average

Advance to Second 0.1 0.5

Advance to Third 0.1 0.6Second to Third 0.0 0.2

Advance to Latest -0.1 1.2Based on the period from 1993 through 2013. The average growth rate in this period (based on today’s data) was 2.6%.

Why So Many Revisions?• Timeliness versus accuracy trade-off.

• “The Bureau emphasized that the … advance estimate … is based on source data that are incomplete or subject to further revision by the source agency”

• Seasonal adjustment.• Seasonal events affecting the economy follow a more or less regular pattern each year.

• Statistics are regularly adjusted to make it easier to observe the longer term movements in the series.

• Conceptual/Definitional Changes.• July 2013: major changes adding in business R&D.

Statistical Significance• Estimates of changes are often not “statistically different” from zero.

• Bureau of Labor Statistics language:• “changed little”• “basically unchanged”

• The estimates may look big.• But, if we can’t say they’re different from zero then we need to look for more information.

• We may not even be sure of the direction of the change!

LABOR MARKET

Employment and Unemployment• Most labor market data are collected by the Bureau of

Labor Statistics (BLS)• Monthly Survey Data

• Current Population Survey (CPS) • A Joint Effort Between the BLS and the Census Bureau• Household survey• Key data: Unemployment Rate

• Current Employment Statistics (CES)• Establishment survey• Eventually “benchmarked” to the “universe” of employed persons• Key data: Payroll employment

• Population = employed + unemployed + not in the labor force

Two Monthly Measures of EmploymentPayrollSurvey

Household Survey

Approx. Sample Size

623,000Individual worksites

60,000 households

“Statistically Significant”

Change114,000 500,000

Demographic Groups? No Yes

Includes self employed, agriculture,

etc?

No Yes

Most Recent Value

(May 2016)

143,894,000Jobs

151,030,000Persons

Employment to Population RatioThe most dramatic graph of the US labor market today

55%

56%

57%

58%

59%

60%

61%

62%

63%

64%

65%

1950 1960 1970 1980 1990 2000 2010

Source: U.S. Bureau of Labor Statistics

May 201659.7%

PRICES AND INFLATION

Prices and Inflation• Overall/Headline Inflation: General rise in the price level.• Consumer Price Index (CPI)

• Produced by the Bureau of Labor Statistics (BLS)• Used to make cost-of-living adjustments to Social Security benefits.

• Personal Consumption Expenditures (PCE) price index • Produced by the Bureau of Economic Analysis (BEA)• As of January, 2012, the Fed has a target of 2% inflation as

measured by the PCE price index.• CPI is generally about 0.5% points higher than PCE

• Average from 1985-2015: CPI is 2.7%, PCE is 2.3%.• Core measures exclude food and energy

• Reduces volatility. • May be a better forecast of future inflation than overall/headline

inflation.

CPI and PCE

PCE and “Core” PCE

FORECASTS

Advocates Love Forecasts• “Marketing salaries for great marketers will double

over the next 5 years.”

• “By 2018, 70 percent of jobs in [Minnesota] will require postsecondary education.”

• “There is not a reason in the world why we cannot grow at a rate of 4 percent a year.”

Economic Forecasts Fail Miserably at Key TimesReal GDP Growth (%)

Real GDP Growth Data SPF One Year Ahead Forecasts Fed Staff One Year Ahead Forecasts

10

8

6

4

2

0

-2

-4

-6

-8

-102000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

Source: U.S. Bureau of Economic Analysis, Philadelphia Fed Survey, and Fed Greenbooks

PCE Inflation Data SPF One Year Ahead Forecasts Fed Staff One Year Ahead Forecasts6

4

2

0

-2

-4

-6

-82000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

PCE Inflation ForecastsPCE Inflation (%)

Source: U.S. Bureau of Economic Analysis, Philadelphia Fed Survey, and Fed Greenbooks

Example: US Employment ForecastsEmployment (in Thousands)

Mid 2009: Overestimated by over 7M jobs!140,000

138,000

136,000

134,000

132,000

130,000

128,0002005 2006 2007 2008 2009 2010 2011 2012 2013

Employment Data 1-Year Ahead Employment Forecast from the Survey of Professional

Forecasters

Source: U.S. Bureau of Statistics, Philadelphia Fed Survey

“Economists put decimal points in their forecasts to show they

have a sense of humor.”- William Gilmore Simms

Are there any good forecasters?•The Federal Reserve Board’s staff is generally the best forecaster for the U.S. economy• But this does not mean good!

•Recently surveys of professional forecasts have mostly caught up to the Fed• e.g. the Philadelphia Fed’s Survey of Professional Forecasters (SPF)

How do we identify bad forecasts?• Biased is bad – in that case they don’t even do well on

average. • Unless focusing only on specific events – then only average over

those events, not full sample.

• Ask how the forecast was made – did they use all available information?

• There can be a lot of luck involved: a bad forecaster can look good for a time, but if you count on them going forward you will get burnt.

• Consider the source of the forecast – is it an advocate?

Good Benchmarks

•Simple models, e.g. historical mean•Averaging across lots of different forecasts

“Trust the weather forecast or always carry an umbrella.”

- Tony Yezer

Macro Pitfalls• Accounting isn’t necessarily causal

• just because GDP = C + I + G + NX doesn’t mean that an increase in C increases GDP

• The economic pie isn’t fixed• Productivity ≠ output• Stocks versus flows

• Deficit and debt are very different things

• Ratios can be trouble

Final Takeaways• Macroeconomic data can send mixed signals as it arrives

in real time.• Macroeconomic forecasts are notoriously unreliable.• When we most want to know the future is when forecasts

are most likely to fail.• So…

• don’t be surprised when forecasts fail.• don’t rely on a single number. • watch the private sector for new data to complement government

data.

THANK YOU!

Tara M. SinclairThe George Washington University

and Indeed

Email: tsinc@gwu.edu

Twitter: @TaraSinc

Contact Information• Full slides from the presentation will be available on my website:

• http://home.gwu.edu/~tsinc/sinclairwc2016.pdf• Feel free to contact me for interviews, background research, or just

curiosity questions about:• Labor market and macroeconomic data and forecasts for the U.S. and around

the world; U.S., Euro-Area, China, Canada, Australia, or global macro or labor market events; Federal Reserve actions; government macro policy or impacts; etc.

• How to reach me:Tara SinclairAssociate Professor of Economics and International AffairsThe George Washington University

Email: tsinc@gwu.edu (generally the best way to reach me)Chief Economist at Indeed

Email: taras@indeed.comFollow me on Twitter @TaraSinc

• Also feel free to check my website for new research:• Web: http://home.gwu.edu/~tsinc/

Data Sources• BEA Website: http://www.bea.gov/index.htm

• Data and methodology for GDP, PCE, etc.• BLS Website: http://www.bls.gov/

• Data and methodology for: CPI, PPI, Employment and Unemployment measures, etc.• Business Cycle Dates: http://www.nber.org/cycles/• Philadelphia Federal Reserve

• Real Time Data: • http://www.philadelphiafed.org/research-and-data/real-time-center/real-time-data/

• Forecasts: • Greenbook data:

• http://www.philadelphiafed.org/research-and-data/real-time-center/greenbook-data/• FOMC forecasts

• http://www.philadelphiafed.org/research-and-data/real-time-center/monetary-policy-projections/

• Survey of Professional Forecasters: • http://www.philadelphiafed.org/research-and-data/real-time-center/survey-of-

professional-forecasters/

More Data Sources• St. Louis Federal Reserve

• Most recently revised data: Federal Reserve Economic Data (FRED): • http://research.stlouisfed.org/fred2/

• “Vintage” data: ArchivaL Federal Reserve Economic Data (ALFRED): • http://alfred.stlouisfed.org/

• Department of Labor’s Unemployment Insurance Weekly Claims Data: http://www.dol.gov/ui/data.pdf

• Cleveland Federal Reserve Inflation Expectations Data: http://www.clevelandfed.org/research/data/inflation_expectations/

• Federal Reserve Board Data Releases: http://www.federalreserve.gov/econresdata/statisticsdata.htm

• Industrial Production and Capacity Utilization• Money Stock and Reserve Balances• Banking, Financial, and Exchange Rate Data

• Divisia Monetary Aggregates from the Center for Financial Stability: http://www.centerforfinancialstability.org/amfm_data.php

• Conference Board Data: https://www.conference-board.org/data/• Consumer Confidence, Help Wanted Online Index, Global Business Cycle Indicators…

Background Research• “The Failure of Forecasts in the Great Recession”

• By Daniel S. Culbertson and Tara M. Sinclair, Challenge. Vol. 57 No. 6 pp. 34-45• http://mesharpe.metapress.com/content/y5772154621m1881/

• GDP Estimates:• Landefeld et al (2008, Journal of Economic Perspectives)

• http://www.bea.gov/about/pdf/jep_spring2008.pdf• BEA’s National Income and Product Accounts (NIPA) Primer

• http://www.bea.gov/national/pdf/nipa_primer.pdf• Federal Reserve Speeches:

• http://federalreserve.gov/newsevents/speech/2012speech.htm• Real time data analysis:

• Croushore (2011, Journal of Economic Literature)• http://www.aeaweb.org/articles.php?doi=10.1257/jel.49.1.72

• Fed’s forecasting ability generally:• Romer and Romer (2000, American Economic Review)

• http://elsa.berkeley.edu/wp/c+dromer_aer2000.pdf• Data Revisions in Recessions:

• Swanson and van Dijk (2006, Journal of Business and Economic Statistics)• http://amstat.tandfonline.com/doi/abs/10.1198/073500105000000036

• Haltom et al (2005, Federal Reserve Bank of Atlanta Economic Review)• http://www.frbatlanta.org/filelegacydocs/erq205_haltom.pdf

• Fed’s forecasting ability in recessions:• Sinclair et al (2010, Economic Letters)

• http://www.sciencedirect.com/science/article/pii/S0165176510001278• Comparing the Fed Staff and the FOMC for forecasting:

• Romer and Romer (2008, American Economic Review)• http://elsa.berkeley.edu/~dromer/papers/aer_98_2.pdf

• “Lucky” Forecasters (including references on averaging forecasts):• Bürgi, Constantin and Tara M. Sinclair. “A Nonparametric Approach to Identifying a Subset of Forecasters that Outperforms the Simple

Average to Empirical Economics.” GW Research Paper on Forecasting Working Paper No. 2015-006.

Other Great Places to Check for Economic Stories• NBER Working Paper Series:

• http://federalreserve.gov/newsevents/speech/2012speech.htm• IGM Experts Panel:

• http://www.igmchicago.org/igm-economic-experts-panel• Shameless Plugs:

• My Website:• http://home.gwu.edu/~tsinc/

• GW Research Program on Forecasting:• http://research.columbian.gwu.edu/cer/research/forecasting

• GW Institute for International Economic Policy:• http://www.gwu.edu/~iiep/

• A few of my favorite blogs:• http://blog.indeed.com/• http://internationaleconpolicy.wordpress.com/• http://econbrowser.com/• http://macroblog.typepad.com/• http://economistsview.typepad.com/• http://marginalrevolution.com/• http://blog.bea.gov/